Hacker News new | past | comments | ask | show | jobs | submit login

I'm reminded of https://www.ba-bamail.com/content.aspx?emailid=24569. Someone please make a similar test of "modern art vs ML-generated art".



I answered 18 correct answers out of 20. Do I qualify as an art critic now?

I assume that the AI generated art would be harder but what is the point really? I would only see the point if the model was trained with something else and ‘it came up with modern art looking images’.


And this works because modern art is flexible enough to allow all kinds of things. Why not do the test on generated text? AI still fails terribly on that mainly because it is not sentient, it does not actually understand anything of it




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: